白烜志
自长江三角洲区域一体化发展上升为国家战略以来，长三角一体化进程不断提速，这就要求交通系统为区域发展提供更为高效的连接效率和更为便捷的出行体验，而这建立在充分理解城市群居民出行规律的基础之上。得益于位置服务技术的发展，大规模居民出行数据的收集成为可能，其中手机信令数据因其覆盖范围广、样本量大、数据稳定可靠等特点，十分适合用来分析城市群居民出行特征。因此，充分利用手机信令数据，挖掘城市群典型出行时空模式，探究影响各模式的显著因素，对于把握城市群出行规律、制定交通政策与交通管理方案、促进区域一体化协同发展具有重要现实意义。
首先，对长三角城市群41个城市5月份的手机信令数据进行预处理与解析。鉴于手机信令原始数据存在的各类问题，通过异常数据识别与处理、基于分箱的规整化方法对手机信令数据进行清洗，使其能够达到交通需求特征分析的要求。然后从OD分布特征、通道分布特征、以及集聚特征3个方面，对长三角城市群的出行特征展开分析。
其次，为进一步探究城市群出行的时空规律，从城市群、都市圈、及核心城市三个层级对城市群出行时空模式进行挖掘。通过手机信令数据根据不同层级的特点生成了出行时空矩阵，使用基于非负矩阵分解（Nonnegative Matrix Factorization, NMF）的方法进行出行时空模式挖掘，得到多种典型出行时空模式及其时空分布。对于城市群出行，以长三角城市群为例，分解出了3种时空模式，如探亲休闲类的往、返程出行和商务类出行。对于都市圈出行，以上海都市圈为例，分解出了4种时空模式，如旅游类、探亲访友类、日常活动类和通勤类出行。对于核心城市内部出行，以上海市为例，分解出了3种时空模式，如早高峰通勤上班类出行、晚高峰通勤归家类出行和购物休闲娱乐类出行。对于核心城市内部出行层级，使用POI数据和地铁刷卡数据，验证了结果的准确性和稳定性。
最后，对城市群出行时空模式的影响因素展开分析。根据对现有研究的梳理，并考虑城际出行产生机理，选取了宏观经济、空间关系、交通便捷、工作生活和旅游休闲五大类影响因素。为更好地反映出行网络所蕴含的城市之间复杂关系，引入社会网络分析领域的QAP模型探究各类因素对3种城市群城际出行时空模式的影响。结果表明，宏观经济、空间关系、出行便捷这三类因素对所有出行时空模式的影响都较大，而工作生活和旅游休闲类因素在不同出行时空模式中的显著性存在明显差异。结果还表明每种出行时空模式的显著因素与其时空分布特征相当契合，这也验证了出行时空模式挖掘方法的有效性。基于建模分析结果，给出了城市群城际出行改进策略和优化案例，并针对毗邻公交发展流程与方法进行了具体的阐述。
本研究利用手机信令数据分析城市群内部出行规律，从多个层次挖掘出行时空模式，并对其影响因素展开分析，最终给出城市群城际出行改进策略，为城市群的更高质量一体化协同提供了一定的发展思路。
关键词：城市群，手机信令数据，出行特征，出行时空模式，影响因素分析
Since the regional integration of the Yangtze River Delta has become a national strategy, the process of Yangtze River Delta integration has accelerated. This requires the transportation system to provide more efficient connection efficiency and more convenient travel experience for regional development, which is based on a full understanding of the travel patterns of urban agglomeration. Benefit by the development of location service technology, collection of large-scale resident travel data has become possible. Among these data, mobile signaling data is very suitable for analyzing the travel characteristics of urban agglomerations due to its wide coverage, large sample size, stability and reliability. Therefore, making full use of mobile signaling data, exploring typical travel spatiotemporal patterns of urban agglomerations, and exploring typical factors affecting of each pattern are of great practical significance for grasping the travel patterns of urban agglomerations, formulating transportation policies and management plans, promoting regional integration and achieving regional coordinated development.
Firstly, preprocess and analyze the mobile signaling data of 41 cities in theYangtze River Delta urban agglommeration in May. In view of the various problems existing in the original data, a series of methods such as map grid processing, abnormal data recognition and processing, spatiotemporal optimizationbased on box splitting method, resident point recognition and travel chain generation are used to clean the mobile signaling data, so that it can meet the requirements of traffic demand feature analysis. Then, analyze the trip characteristics of the Yangtze River Delta urban agglomeration from 3 aspects: OD distribution characteristics, channel distribution characteristics, and agglomeration characteristics.
Secondly, in order to further explore the spatiotemporal patterns of urban agglomeration travel, the spatiotemporal patterns of urban agglomeration travel are mined at different levels. Based on the characteristics of different levels, use mobile signaling data to generate travel spatiotemporal matrix. Various typical travel spatiotemporal patterns and their spatiotemporal distributions are obtained by using the non negativematrix factorization (NMF) method to mine travel spatiotemporal patterns. For urban agglomeration level, taking the Yangtze River Delta urban agglomeration as an example, three spatiotemporal modes were decomposed, such as the go and back travel of tourism, home and friend travel, and business commuting travel. For metropolitan area level, taking the Shanghai metropolitan area as an example, four spatiotemporal patterns were decomposed, such as the travel of tourism and leisure, visiting family and friends, daily activities, and commuting business. For core city level, taking Shanghai as an example, three spatiotemporal patterns were decomposed, such as the travel of commuting to work at morning peak, commuting back home at evening peak, and daily activities. For core city level, the accuracy and stability of the results were verified by using POI data and subway data.
Finally, analyze the influencing factors of the spatiotemporal patterns of urban agglomeration travel. Based on the analysis of existing research and the mechanism of intercity travel, five major influencing factors were selected: macroeconomic, spatial relationships, convenience of transportation system, work life, and tourism leisure. To better reflect the complex relationships between cities contained in the travel network, QAP model in the field of social network analysis is led to explore the impact of various factors on the three spatiotemporal patterns of urban agglomeration intercity travel. The results indicate that macroeconomic, spatial relationships and convenience of travel have a significant impact on all travel spatiotemporal patterns, while work life and tourism leisure have significant differences in their significance among different travel spatiotemporal patterns. The results also indicate that the significant factors of each travel spatiotemporal pattern are quite consistent with their spatiotemporal distribution characteristics, which also verifies the effectiveness of the travel spatiotemporal pattern mining method. Based on the modeling analysis results, improvement strategies and optimization cases for intercity travel in urban agglomerations were also presented, and specific explanations were provided for the development process and methods of adjacent public transportation.
This study uses mobile signaling data to analyze the internal travel patterns of urban agglomerations, excavates travel spatiotemporal patterns from multiple levels. and analyzes their influencing factors. Finally, an improvement strategy for intercity travel in urban agglomerations is proposed, providing a certain development idea for the high-quality integrated and coordinated development of urban agglomerations.
Key Words: urban agglomerations, mobile signaling data, travel characteristics, travel spatiotemporal patterns, influencing factors analysis
